How to Improve DALL-E Prompts (Tips & Examples)
Hi! This tutorial will show you how to enhance your DALL-E prompts for better AI image generation.
In a previous tutorial, we discussed best DALL-E prompts, and saw some example images generated using these prompts.
In this one, we will examine specific ways we can improve DALL-E prompts in order to get better image generation results.
To follow along, you can either log in to the Open AI DALL-E lab or download and install DALL-E desktop app on your computer.
Here is an overview:
Let’s get into the discussion!
Tip 1: Be Detailed in Your Prompt Description
DALL-E is an AI that has been trained on hundreds of millions of images. Therefore, it knows a lot about many things, especially as it is also constantly fine-tuned.
This means that you can give it a simple one-sentence prompt, and it will generate the image for you.
Nevertheless, it is recommended that you should be detailed in your prompt description so that DALL-E knows your goal with regard to the image and the features you want in the image.
Let’s now contrast the results of a simple prompt and a detailed prompt:
Prompt: “A robot playing guitar”
As you can see in the above image, we get a very simplistic result. But what happens if we add some detail to the prompt? Let’s see.
Detailed prompt: “a 3D, neon, futuristic, mechanical robot playing an electric guitar”
Now compare the above image with the previous one. You can see a vast contrast between both images.
By just adding some detail to the original prompt, DALL-E has generated a better image than the previous one.
Tip 2: Pay Attention to the Image Background
Another way to improve a DALL-E prompt is to be specific about the background of the image. As you might have noticed, with DALL-E images, they tend to have a generic blank background.
However, we can specify the kind of background we want an image to have by including that in the prompt.
We will add that detail to the robot playing an electric guitar prompt from the previous example and see what result DALL-E will return.
Prompt: “a 3D, neon, futuristic, mechanical robot playing an electric guitar in a stadium”
Now we see the robot playing an electric guitar with the backdrop of a stadium. Note that this image is quite different from the one in the last example. This is because of the random nature of generative AI.
As many times as you run the prompt, as many times as you will get different images, although within the same context.
Tip 3: Specify the Art Style
You can also improve the output of your DALL-E prompt by specifying an art style for the image. For instance, you could include in the prompt that you want the image to have a cartoonish style, an artwork style, abstract style, or even a gothic style.
Let us do that to our former prompt, and see what DALL-E outputs.
Prompt: “Pixar style 3D, neon, futuristic, mechanical robot playing an electric guitar in a stadium”
If you are familiar with Pixar Animation Studios kind of cartoons, you will agree that the above image really resembles a Pixar production.
You can think of other ways to style the output of the DALL-E prompt.
Tip 4: Be Specific About Camera Angles
Another tip to improve your DALL-E prompt is to specify the camera angle of the image. We will demonstrate 2 camera angles: a close-up shot and an aerial shot.
Prompt: “close-up shot of a Pixar style 3D, neon, futuristic, mechanical robot playing an electric guitar in a stadium”
Prompt: “aerial shot of a Pixar style 3D, neon, futuristic, mechanical robot playing an electric guitar in a stadium:
Clearly, by specifying a camera angle in your input, you can achieve a unique effect.
Tip 5: Refrain from having more than One Character in the Image
This tip might be subjective, but experience shows that it can be quite challenging to get DALL-E to generate an image comprising two characters in a really coherent and distinct manner.
Let’s see an example using our robot prompt.
Prompt: “Pixar style 3D, neon, futuristic, mechanical robot and cat playing an electric guitar in a stadium”
As you can see, DALL-E didn’t quite return the desired output. While we wanted an image comprising a robot and a cat, both separately playing electric guitars, we rather got an image where the robot has been merged with the cat and playing an electric guitar.
However, this is not to say that our goal cannot be achieved with DALL-E. What it ultimately boils down to is spending time tweaking your input until DALL-E eventually gives you what you want.
In conclusion, there are so many other ways you can enhance your DALL-E prompts, such as by specifying the lighting of the image, specifying the setting of the image, and even editing the image.
What is required is patience and quite some creativity and imagination. In the end, DALL-E is just a robot that needs to be told what to do.
Video, Further Resources & Summary
Do you need more explanations on how to enhance your DALL-E prompts? Then you should have a look at the following YouTube video of the Statistics Globe YouTube channel.
In the video, we explain in some more detail how to enhance your DALL-E prompts.
The YouTube video will be added soon.
Furthermore, you could have a look at some of the other interesting AI-based tutorials on Statistics Globe:
- How to Make AI-Generated Art Using Python (Example)
- What is Image Classification?
- Image Classification Using Hugging Face transformers pipeline in Python (Example)
- How to Download & Install DALL-E App
In this tutorial, we have explained 5 tips for enhancing your DALL-E prompts along with examples. I certainly hope you found it helpful!
In case you have further questions, you may leave a comment below.
This page was created in collaboration with Ifeanyi Idiaye. You might check out Ifeanyi’s personal author page to read more about his academic background and the other articles he has written for the Statistics Globe website.
Statistics Globe Newsletter